Preterm delivery (PTD) is the leading cause of infant morbidity and mortality in the United States, and prevention is a primary goal for perinatal health care. Recent evidence indicates that there is a strong association between an altered vaginal microflora during pregnancy and the occurrence of PTD. However, the role of specific microorganisms in PTD is not well understood. Recent studies in this laboratory have been directed at identifying specific microbial risk factors associated with PTD. We have established a predictive statistical model for PTD, based on these microbiologic risk factors. Our studies have identified two key populations that are associated with the occurrence of PTD. It has been shown that the levels of a bacterial phospholipase, PLA2, increase in concentration between 20 and 30 weeks of gestation in women who deliver at less than 37 weeks gestation. This increase in PLA2 correlates with the presence of Prevotella sp. as part of the vaginal microflora. It has also been noted that the presence of both hydrogen peroxide (H2O2) producing lactobacilli and non-H2O2 producing lactobacilli simultaneously as part of the vaginal microflora is a risk factor for PTD, separate from the presence of either strain by itself. This observation suggests that combinations of strains may have a detrimental effect on pregnancy outcome. The goal of the proposed project is to prospectively collect quantitative and qualitative microbiologic data from women at high risk for PTD and those with no identifiable risk for PTD, to determine whether the statistical model is capable of predicting PTD based on microbiologic data, in preparation for a subsequent study in which women identified as at risk for PTD by this model will be treated. In addition, the actual mechanism(s) by which bacteria cause PTD will be evaluated in order to identify microbiologic targets for any interventional study. The specific aims for this proposal are to 1) refine the predictive model derived from the initial analysis of the data collected during the first three years of this study and to validate this model for predicting PID during a prospective clinical trial, 2) to improve out understanding of the role of specific bacterial species in PTD particularly Prevotella sp, 3) to use molecular typing methods to determine whether specific strains of lactobacilli are more common in women delivering at less than 37 weeks gestation than those delivering at 37+ weeks gestation and whether such strains are acquired during pregnancy, and 4) to determine whether interactions between strains of lactobacilli deleterious to pregnant women occur by using cultured vaginal epithelial cells in vitro exposed to combinations of lactobacilli found in vivo.